Diana L. MacLean

MoodWings: a wearable biofeedback device for real-time stress intervention

Stress has a wide range of negative impacts on people, ranging from declines in real-time task performance to development of chronic health conditions. Despite the increasing availability of sensors and methods for detecting stress, little work has focused on automated stress interventions and their effect. We present MoodWings: a wearable butterfly that mirrors a users real-time stress state through actuated wing motion. We designed MoodWings to function both as an early-stress-warning system as well as a physical interface through which users could manipulate their affective state. Accordingly, we hypothesized that MoodWings would help users both calm down and perform better during stressful tasks. We tested our hypotheses on a common stressful task: driving. While users drove significantly more safely with MoodWings, they experienced higher stress levels (physiologically and self-perceived). Despite this, users were enthusiastic about MoodWings, expressing several alternative contexts in which they would find it useful. We discuss these results and future design implications for building externalized manifestations of real-time affective state.

Identifying medical terms in patient-authored text: a crowdsourcing-based approach

Background and objective As people increasingly engage in online health-seeking behavior and contribute to health-oriented websites, the volume of medical text authored by patients and other medical novices grows rapidly. However, we lack an effective method for automatically identifying medical terms in patient-authored text (PAT). We demonstrate that crowdsourcing PAT medical term identification tasks to non-experts is a viable method for creating large, accurately-labeled PAT datasets; moreover, such datasets can be used to train classifiers that outperform existing medical term identification tools. Materials and methods To evaluate the viability of using non-expert crowds to label PAT, we compare expert (registered nurses) and non-expert (Amazon Mechanical Turk workers; Turkers) responses to a PAT medical term identification task. Next, we build a crowd-labeled dataset comprising 10 000 sentences from MedHelp. We train two models on this dataset and evaluate their performance, as well as that of MetaMap, Open Biomedical Annotator (OBA), and NaCTeMs TerMINE, against two gold standard datasets: one from MedHelp and the other from CureTogether. Results When aggregated according to a corroborative voting policy, Turker responses predict expert responses with an F1 score of 84%. A conditional random field (CRF) trained on 10 000 crowd-labeled MedHelp sentences achieves an F1 score of 78% against the CureTogether gold standard, widely outperforming OBA (47%), TerMINE (43%), and MetaMap (39%). A failure analysis of the CRF suggests that misclassified terms are likely to be either generic or rare. Conclusions Our results show that combining statistical models sensitive to sentence-level context with crowd-labeled data is a scalable and effective technique for automatically identifying medical terms in PAT.

Designing a prototype interface for visual communication of pain

Thousands of people use Online Health Communities (OHCs) as a forum for expressing and collaborating on symptoms of pain. Despite the physical nature of pain, these exchanges typically comprise text. While pain referral diagrams have served as patient-physician communication aids for decades, little research has focused on translating them into an interactive digital interface. We propose that such an interface would provide a more efficient and accurate mechanism for expressing pain and would facilitate useful discussion around pain symptoms. In this work-in-progress, we present a pilot study in which users expressed physical symptoms using pen and paper. Our results uncovered several design considerations that are currently being used to inform the design of Body Diagrams, an interactive pain visualization tool that we plan to deploy to a pain-related OHC in the near future.

On Recovering Structure of Affect

This paper presents novel human computation experiments geared towards uncovering the structure of affect. Using Mechanical Turk workers across 2 separate studies, we empirically verified some of the popular beliefs about the structure of affect, but also provide some new evidence. We replicate and reveal not only the statistical structure of the dimensions of affect, but also the effect of cultural influences. We close with a proposition for a framework for doing this kind of large scale research and provide recommendations and opportunities for innovations in research around emotional theory.

Gathering people to gather data

An interview with Paul Wicks, Vice President of Innovation at PatientsLikeMe, a patient network and real-time research platform.